An Event-triggering Approach for Bus Tracking based on Multimodel Mobility Prediction

Abstract

Tracking a fleet of buses that serve routes across a city is an essential functionality for the successful utilization of modern transportation networks. However, to date, little attention has been paid to the effectiveness of the implemented tracking algorithms, which typically rely on periodic signaling messages sent by the buses in motion to indicate the vehicle location. However, as we will demonstrate in this work the latter approach entails a large number of unnecessary signaling messages to be communicated in order to achieve a high level of tracking accuracy.The alternative approach we present in this work is based on a novel event-triggering strategy that substantially improves tracking and reduces significantly the number of messages that need to be sent out compared to periodic signaling. Time-series traces are used to extract and update mobility models which are then used to estimate future vehicle locations. The estimate, and actual arrival times at predetermined locations are assessed by onboard units and an event is triggered whenever the deviation exceeds the desired tracking accuracy.